►
From YouTube: KubeEdge Community Meeting 20200527 (Europe Time)
Description
No description was provided for this meeting.
If this is YOUR meeting, an easy way to fix this is to add a description to your video, wherever mtngs.io found it (probably YouTube).
A
Okay,
the
recording
is
on.
Let
me
share
the
agenda.
A
All
right,
so
thank
you.
A
Thank
you,
everybody
for
joining
the
meeting
today.
It's
may
27th
may
27th
2020.
Today
is
europe
time
friendly,
qh
community
meeting.
A
So
we
have
a
few
items
on
the
agenda
and
let
me
check
if
there's
a
new
attendee,
maybe
we
can
ask
them
to
briefly
introduce
themselves,
and
I
see
that
king
is
raising
his
hand.
King,
would
you
like
to
introduce
yourself.
A
A
Okay,
channeling
here,
would
you
like
to
introduce
yourself.
C
Oh,
oh
okay,
I
have
talked
which
I
want
to
share
friends
like,
and
I
just
in
this
meeting.
I
want
to
share
that.
I'm
working
on
considering
building
robotics
is
attracted
to
the
system
based
on
kubernetes,
also
kbh.
A
C
A
Sure
I
will
introduce
invite
the
others
to
introduce
themselves
and
then
we
can
move
on
to
the
agenda.
Okay,
okay,
thank
you.
Davinci.
Would
you
like
to
introduce
yourself.
A
A
Okay,
diving
is
not
so
convenient,
so
maybe
we
can
just
move
on.
A
All
right,
if
anyone
would
like
to
introduce
yourself,
please
try,
try
me.
Otherwise
we
can
move
on
to
the
agenda.
A
Yeah
or
you
can
feel
free
to
introduce
when,
when
you
want
to
give
comments
all
right,
so
let's
move
on
to
the
agenda
today.
The
first
would
be
introduction
from
trending
here.
A
C
Oh
sorry,
no,
my
my
phone
was
in
this
meeting.
C
C
A
Should
I
play
the
slide
or
you
can
just
look.
C
A
Okay,
so
I'm
now
playing
the
slide.
If
you
need
to
move
on
to
the
next
page,
just
feel
free
to
tell
me.
A
C
A
C
C
A
Yes,
yes,
should
I
move
on
next
or.
C
C
B
A
So
so,
just
let
me
briefly
introduce
some
of
the
attendees
prefer
to
speak
in
chinese,
so
I
will
have
to
translate
okay,
okay,
thank
you.
C
C
A
Okay,
so
so
a
brief
about
the
background,
so
they
are
trying
to
build
a
distributed
system
based
on
4d
robotics.
So
the
challenge
is
that
the
robotics
system
itself
is
not
distributed
right.
The
real
right
this
is
the
name
of
the
system,
is
called
the
rules
and
they
think
that
so
actually
kubernetes
did
a
lot
of
good
things
on
running
the
distributed
things,
but
the
challenges
that
encounter.
A
A
Is
that
running
kubernetes
on
arm
based
architecture,
so
even
for
cubelet
itself,
it
consumes
a
lot
of
memory.
That's
one.
A
So
the
second
challenge
is
that
the
in
real
world
use
case
the
applications
may
not
run
in
the
same
network,
not
in
the
same
land,
but
kubernetes
need
all
the
components
running
same
length.
So
that's
the
second
challenge.
C
Oh
yes,
this
angle,
integer,
should
not
know,
may
go
to
the
end
and
put
your
appearance
insurance.
A
Okay,
the
third
challenge
is
that
for
the
no
no,
the
component
cubelet
are
not
able
to
not
able
to
run
or
turn
the
atom
in
autonomously
when
it's
disconnected
right.
C
A
A
Okay,
so
so
they
found
that
for
cuba,
the
the
design
matching
their
requirements
quite
well,
but
there
are
still
some
challenges
when
adopting
on
cuevage.
So
the
major
one
is
that
the
applications
they
run
on
each
node.
They
need
to
discover
each
other
by
broadcasting
and
currently
the
year
fund
solution
to
do
that
right.
C
C
A
Okay,
so
they
tested
that
the
ci
wave
can
support
their
case.
A
Okay,
so
they're
currently
still
doing
more
detailed
research
to
to
try
to
run
the
whole
thing
with
curvature.
Okay,
okay,
thanks
all
right,
so
that's
all
about
the
the
first
thing.
So
actually
I
can
also
paste
the
link
to
the
agenda,
so
everyone
can
see
that.
A
A
All
right,
so
I
I
can
give
some
background
about
this,
so
we
know
that
we
have
a.
B
A
Ai
offloading
a
blueprint
so
in
the
other
meeting
today,
they
are
discussing
about
so
basically
designing
the
layer
for
ml
lending
offloading
framework
built
on
top
of
qh,
and
they
want
to
clarify
that,
unlike
how
many
apis
expose
the
exposed
from
cuba,
age,
they
can
consume,
and
then
they
can
better
provide
the
framework
for
running
for
supporting
different
machine
learning,
ai
framework
frameworks
on
top
of
the
their
offloading
layer.
I
think
that's
the
background
so
from
cube
edges
itself,
the
community-wide
we.
A
I
think
that
currently
our
document
is
not
enough.
This
is
also
this
also
have
been
mentioned
multiple
times.
We
need
a
api
reference
to
describe
what
kind
of
apis
are
provided
by
qh
and
what
are
the
details
about
the
of
each
field?
Okay,.
A
All
right,
the
next,
I
think
we
should
also
go
through
the
1.4
release,
planning.
A
Okay,
so
there
are,
some
change
need
to
be
done
in
upstream
kubernetes
about
the
certification
thing.
So
currently,
changi
is
working
on
that,
after
that
he
will
start
working
on
the
exec
all
right.
A
Next
metric
server.
A
Okay,
so
any
updates
on
the
golan
upgrade.
I
think
we
are
still
pending
right.
A
A
E
A
Okay,
if
not
from
now,
we
can
just
leave
here,
please
feel
free
to
add
your
name.
If
you
want
to
to
take
this
item.
A
B
A
A
So
next
is
about
the
device
crd
improvement.
Currently,
the
the
crd.
The
proposal
is
still
under
review.
Please
everyone-
if
you
are
interested
in
in
that,
please
take
a
look
and
leave
a
comment
there.
Okay,.
A
Okay
for
the
hq
running
age
core
as
a
systems
process
process,
I
think
this
one
is-
is
ready
to
merge.
Okay,
next
debian
packages
for
cuevage,
who
is
working
on.
E
A
A
Anyone,
okay,
so
I
think
yeah
vj
is
working
on
that
right.
A
Okay,
maybe
we
we
can
get
updates
from
a
vj
later
on.
A
Next
edu,
you
know
cross
subnet
communication.
I
think
that
was
still
the
discussion,
no
no
proposal.
Yet.
Okay
and
the
next
three
also
no
proposal,
yet
all
right.
So
that's
more
about
the
1.4
updates,
so
something
else
is
about
so
we
just
started
preparing
for
the
for
the,
including
incubation
application
I
submitted
the
presentation
slide
to
since
they
have
seen
runtime
last
week,
but
not
yet
get
chance
to
do
the
presentation,
so
that
will
be
probably
done
in
the
next
week.
A
So
yeah
they
run
our
meetings
every
two
weeks
and
and
in
the
same
in
the
meantime,
the
sick,
seaguar
missy
each
other
and
the
sig
device
charter
is
still
on
the
discussion.
A
Not
okay,
we
we
did.
We
did
some
updates
on
the
charter,
so
maybe
I
just
quickly
go
through
it
and
if
anyone
are
interested
in
this,
please
take
a
look
and
please
take
a
look
and
leave
your
comments.
Okay,.
A
So
for
the
for
the
device
iot,
as
as
we
introduced
before,
it's
more
to
focus,
improving
the
focus
on
improving
the
device
crd,
as
well
as
the
mapper
architecture,
to
make
it
easy
to
integrate
with
more
industrial
devices,
including
multiple
device
device
communication
protocols,
as
well
as
make
it
able
to
represent
more
type
of
devices.
A
So
this
the
errors
of
folks
for
this
scope
for
this
thing
would
be
defining
the
common
api
abstraction.
So,
basically,
the
the
device
crds,
as
well
as
improve
the
mapper
and
the
device
screen
framework
to
make
it
able
to.
A
Integrate
with
the
devices
in
different
scenarios
like
the
some
of
the
device,
maybe
just
the
switch
and
some
of
the
sensors,
they
get
a
lot
of
data,
and
also
we
may
in
this.
In
this
thing,
we
need
to
focus
on
simplifying
the
application
development
to
integrate
with
the
device.
So
we
we
may
provide
a
reference
implementation,
so
basically
the
mappers-
and
maybe
we
can
also
provide
some
of
the
code
skeleton
for
for
end
user
to
develop
new
mappers.
A
And
also
for
cross-community
collaboration,
this
egg
is
respon,
will
be
responsible
to
to
investigate
and
probably
provide
a
tooling
to
to
do
some
verification
and
maybe
later
on,
even
promote
some
conformance
testing
or
promote
some
standard.
Okay
also
and
the
data
management
on
the
age.
So
data
management
management
here
is
more
more
likely
means
the
data
produced
by
the
device.
A
So
this
would
will
also
focus
on
defining
the
process
or
workflow
will
make
it
possible
for
end
user
to
define
workflow
for
processing
the
data
and
age
and
the
regenerate
like
warning
or
some
events.
A
Okay,
so
that's
basically
so
far,
the
updates
about
basic
and
for
amazing
mec.
There
are
actually
a
lot
of
discussion
ongoing
and
for
the
scope.
Currently,
we
divided
to
three
major
aspects:
one
is
for
one
is
for
mec
services,
so
basically
there's
some
relevant
functionality
to
to
enabling
the
mvc
whole
architecture.
A
So
the
upf,
the
user
plan
functionality,
whether
it's
it
should
be
covered
in
cubase
or
not
is
still
under
discussion,
probably
will
not
for
a
networking
the
basic.
Basically,
this
thing
will
need
to
figure
out
how
to
how
container
networking
is
required
in
this.
A
For
the
mec
and
if
there's
any
functionality
to
be
improved
like
security,
multi-tenancy
or
network
isolation
or
bandwidth,
isolation,
sort
of
things-
and
the
third
aspect
is
the
mec
infrastructure,
so
that
one
includes
like
integrating
with
the
hardware
accelerator
and
also
the
multiple
kubernetes
cluster
management,
including
qh.
A
So
basically,
the
the
this
is
responsible
for
providing
a
reference
architecture
using
curvage
to
enable
mec
and
make
it
easier
for
the
technical
operators,
as
well
as
the
ecosystem,
vendors,
to
develop
mec,
relevant
framework
functionalities,
business
on
the
reference
architecture,
and
also
this
sig
will
also
investigate
the
the
cross
community
collaboration
and
also
probably
provide
some
testing
and
verification
tools.
A
A
A
So
if
you
are
interested
in,
please
take
a
look
and
leave
your
comments.
There.
A
All
right,
that's
more
about
the
items
on
the
agenda
today,
so
we
have
some
minutes
free.
Anyone
would
like
to
discuss
anything.
Please
feel
free
to
join
me.
F
Kevin,
I
have
a
question
about
this
six,
so
yeah.
If
we
want
to
create
this
thing,
should
we
answer
the
permission
from
the
cncf
or
ask
permission
from
some
other
organization,
or
we
can
just.
A
Manage
it
by
ourselves,
so
these
things
just
to
be
clear,
this
thing
is,
is
the
cube
h6,
so
it's
okay
under
cubase,
we
we
can
manage
things.
Managing
I
mean
we
can
make
the
decision
inside
cuba
itself.
We
don't
need
any
approval
from
the
other
organization.
B
F
Interested
in
summer,
one
of
these
six,
so
he
can
just
maybe
he
can
put
his
names
or
anywhere
or
or
I
mean
how
to
how
to
apply
to
join
this
thing.
A
Well,
this
thing
is
just
like
the
way
we
hold
the
community
meeting
so
so.
Currently
the
the
the
sig
already
have
the
slack
channel
created,
so
everyone
is
able
to
join
that
slack
channel,
and
so
the
one
thing
not
yet
confirmed
is
that
each
sig
do
not
have
a
fixed
meeting
time.
A
So
so
currently
the
discussion
will
be
the
discussion.
Time
will
be
confirmed
inside
the
slack
channel.
So
please
join
the
slab
channel
too,
to
keep
updated
and
to
formally
make
the
sig
running
we
need
to.
We
need
to
find
a
fix
the
time
for
each
sig
meeting
and
also
we
need
to
finalize
the
charter.
A
All
right
any
other
ideas
or
questions.
A
E
A
A
If
there's
any
plan
about
doing
the
test,
stability
checking
or
chaos
testing,
so
actually,
actually
we
have
some
offline
discussion
with
the
other
project
maintainer.
A
So
maybe
one
of
you
know
about
the
one
project
called
chaos
mesh,
so
chaos
mesh
is
one
of
the
two
using
service
mesh
technology
to
do
chaos,
engineering
on
top
of
kubernetes.
A
We
have
some
initial
idea
to
using
chaos
mesh
to
do
chaos,
testing
for
2h
to
verify
the
this
kind
of
disconnected
scenario,
as
well
as
like
engineered
offline
or
or
package
loss,
or
a
high
lengthy
network
environment
and
but
the
it's
still
under
discussion,
and
we
do
need
people
to
volunteer
joining
network
to
to
make
it
happen.
So
for
the
framework
we
actually
have
one
testing
framework.
So
that's!
A
The
idea
was
to
run
cubase
on
a
kubernetes
cluster,
so
basically
using
kubernetes
to
replace
the
virtualization
thing
and
you
then
you
don't
need
the
create
multiple
nodes
on
any
of
the
public
cloud
or
on
your
laptop.
You
can
just
run
simulate
the
whole
cubase
cluster
with
multiple
hd
nodes
on
one
of
the
kubernetes
cluster.
A
So,
in
that
case,
we
can
easily
enforce
chaos
mesh
to
to
do
to
simulate
some
simulate
some
broken
environment
and
verified
the
curvature
functionality.
A
Okay,
that's
that's!
That's
the
the
idea
and
what
and
has
been
discussed
so
far,
and
we
need
some
volunteers
to
join
that
work.
A
E
E
E
A
F
A
Foreign,
so
the
question
is
that
one
of
the
corner
cases
that
when,
when
one
h
note
become
offline
and
the
disk
may
be
broken
so
the
so
basically
or
the
circular
file
broken
then
when
it
come
back,
they
it
needs.
A
It
need
to
fetch
all
the
node
reference
relevant
data
from
cloud.
Well,
currently,
the
status
like
recorded
in
the
crd
and
the
sync
and
the
cluster
sync
in
in
the
cloud
the
they
think
that
all
the
metadata
is
up
to
date
on
the
agenda.
A
A
E
D
A
Foreign,
that's
all
thanks,
okay,
so
the
two
questions
just
to
discuss
these
one
of
the
one
is
that
the
when
we
are
implementing
the
leader
election
for
cloud
core,
because
we
are
reusing
the
functionality
provided
by
kubernetes.
A
So
the
that
mechanism
will
update,
keep
a
periodically
updated
endpoint
of
the
cloud
core
instance
and
because
we
also
are
providing
the
service
functionalities
at
age.
So
the
end
of
one
any
endpoint
updates
would
be
synced
to
the
age.
So,
but
actually,
the
updates
of
cloud
core
instance
is
is
not
quite
necessary
from
the
age.
A
A
We
can
add
some
filtering
mechanism
to
to
make
only
useful
updates
synchronized
to
the
age
and
also
another
relevant
discussion
about
this-
that,
for
the
whole
service,
endpoint
updates,
we
can
also
figure
out
maybe
for
the
updates
of
end
point
that
are
located
in
the
cloud.
We
don't
need
to
sync
all
the
things
to
the
edge.
We
can
roughly
make
the
make
it
able
to
distinguish
when
the
application
is
accessing
the
endpoints
in
the
cloud
and
just
simply
route
the
connected
to
the
20
cloud.
A
For
all
the
end
of
the
end
points
on
the
age,
we
may
still
need
to
to
sink
all
the
things
to
the
edge.
So
that's
just
some
rough
idea.
We
can
figure
out
more
details,
and
probably
there
will
be
some
feature
enhancement
in
these
cases.
A
G
F
A
A
Okay,
so
we
are
running
out
of
time.
Thank
you,
everybody
for
joining.
We
discussed
a
lot
of
things
today,
so
the
the
meeting
next
time
would
be
july
3rd.
It
will
be
pacific
time
next
week.